jdoerfert

8 ms·

1.
by jdoerfert 3y ago
This is not true. 3. is supported by AMD on new hardware, e.g., Frontier. See https://docs.olcf.ornl.gov/systems/frontier_user_guide.html#...
2.
by jdoerfert 3y ago
Yes, though with caveats. The driver and parts of the extended API we used to lower CUDA calls are in upstream LLVM. The wrapper headers are not. We will continue the process of getting it all to work in upstream/vanilla LLVM soon thou
3.
by jdoerfert 3y ago
Shameless plug: https://www.osti.gov/servlets/purl/1892137 TLDR; If you provide even more functions through the overloaded headers, incl. "hidden ones", e.g., `__cudaPushCallConfiguration`, you can use L
4.
by jdoerfert 6y ago
Generally, yes. Though, as always, these things are hard to summarize in one sentence and depend on the standard version (pre/post C++11). This post might shed some light on the options we are considering for Clang right now: https:&#
5.
by jdoerfert 6y ago
Since https://reviews.llvm.org/D86841 , ~ Aug 2020, Clang is able to reason about the forward progress guarantees of the input. For the LLVM-IR handling see https://reviews.llvm.org/D86233 . While we are stil
6.
by jdoerfert 6y ago
FWIW, `noalias`, aka. `__restrict__`, generally works fine but the inliner translates it into `noalias` metadata which is, still, ambiguous and also not handled properly in some cases. Loop unrolling is one of them but that will be fixed wi
7.
by jdoerfert 7y ago
Does this sound interesting: http://llvm.org/docs/ProgrammersManual.html#llvm-adt-indexed... http://llvm.org/doxygen/IndexedMap_8h_source.html
8.
by jdoerfert 7y ago
Clang will not "reduce compilation effort" and I do not believe GCC does that. There are heuristics that look at things like function size but I don't think that is what they mean. Generally speaking, translation units are in
9.
by jdoerfert 7y ago
The godbolt compiler explorer, which is awesome btw, had to tackle all sorts of problems like this. I think there is a talk on YouTube that goes into some detail.
10.
by jdoerfert 7y ago
It would be news to me that the PolyhedralInfo interface is actually used anywhere. As the file comment says, this was "work in progress" a few years ago and it shares the same drawbacks as Polly (unfortunately). Polyhedral develo